Read aloud

A voice reads the article in order, and the word it is on is highlighted as it goes. Browsers and operating systems ship it under several names: Read Aloud, Speak Selection, text to speech, and the travelling highlight is often called karaoke highlighting, which is exactly what it looks like. The whole feature is one sentence long: the prose, out loud, with your place kept.

The reason it needs its own word is that it keeps getting confused with a screen reader, and the two do different jobs. A screen reader is how somebody who cannot see the page operates it: it announces structure, roles and state, so a reader hears “heading level two”, “button, collapsed”, “list, six items”, and navigates by those. Read aloud announces none of that. It voices the running text and stops there, for a reader who can see the screen and would rather have the words carried. Mixing them up leads teams to build the wrong thing, and the expensive version of that mistake is a “listen to this page” button shipped in place of accessible names, roles and focus order.

Who it is for is broader than most accessibility features: dyslexic readers, readers with attention difficulties, low vision readers who use it alongside magnification, people reading in a second language, and anybody whose hands or eyes are busy. Two design consequences follow. Reading order is spoken order, so furniture that sits in the middle of the flow (a share bar, a related list, a pull quote repeated from below) gets read as part of the article unless it is moved out of it. And prose written for assistive technology is prose here too: a long description under a chart is read out like any other paragraph, which is a good reason to write it as prose rather than as a list of coordinates.

If you are building the control rather than relying on the platform’s, the parts that matter are play and pause, a speed setting, a voice setting, and the word highlight itself, because the highlight is what lets a reader who drifted find their place again without rewinding. Keep it clear of text selection, since a reader following along often wants to select the sentence they just heard. Most products should let the browser and the operating system do this and spend the effort on the markup underneath, which both features read from.